Skip to content

Latest commit

 

History

History
33 lines (24 loc) · 1.04 KB

README.md

File metadata and controls

33 lines (24 loc) · 1.04 KB
Raw

Neural Causation Coefficient

An unofficial PyTorch implementation of Neural Causation Coefficient (NCC) in the paper, "Discovering Causal Signals in Images".

Requirements

torch=1.9.0
torchvision=1.10.0
scikit-learn
scipy
tqdm
numpy
pandas
matplotlib
seaborn
skimage
pycocotools

Usage

  1. Download datasets: 1) Tubingen dataset, 2) PASCAL VOC2012 dataset, and 3) MSCOCO dataset.

  2. NCC_Training.ipynb: Train NCC Classifier on synthetic dataset and Test it on Tubingen dataset

    • Our reproduced model results/NCC_classifier_best.pt showed 78.74% accuracy on Tubingen dataset, which was 79% in the original paper.

  3. ObjectClassifier_Training.ipynb: Train object classifier on PASCAL VOC2012 dataset

  4. NCC_Testing.ipynb: Test NCC Classifier on MSCOCO dataset

References

David Lopez-Pas et al., Discovering Causal Signals in Images, ICML 2017